From Models to Agents: Building Context-Aware Consumer AI at Scale at DoorDash

Consumer AI is rapidly evolving beyond one-shot predictions toward systems that reason, plan, and adapt in real time. In this talk, Sudeep explores how personalization at DoorDash is shifting from static recommendation models to agentic, goal-driven AI experiences that operate across search, discovery, decision-making, and fulfillment. Using real examples inspired by DoorDash’s marketplace, and consumer surfaces, the session dives into how large language models, deep learning, and reinforcement-style feedback loops can be combined to build consumer-facing agents that understand intent, maintain context across sessions, and take coordinated actions—from helping users find the right meal to optimizing trade-offs across price, speed, quality, and reliability. Topics include agentic product discovery in a dense marketplace, multi-step decision flows that span browsing to checkout, building trust and controllability into consumer agents, and practical evaluation strategies for agentic systems operating at DoorDash scale. Attendees will leave with concrete frameworks for designing AI systems that feel helpful, adaptive, and human-centered—while remaining reliable, efficient, and grounded in real-world marketplace and business constraints.